931

A Comparison of High-Performance Liquid Chromatography and Spectrophotometry to Measure Chlorophyll in Canola Seed and Oil Kerry Ward a,*, Rachael Scarth a, J.K. Daun b and C.T. Thorsteinson b aDepartment of Plant Science, University of Manitoba, Winnipeg, Manitoba R3T 2N2, Canada and bGrain Research Laboratory, Canadian Grain Commission, Winnipeg, Manitoba R3C 3G8, Canada

There are several methods available to measure chlorophyll in eanola oil and seed, and these will not necessarily yield the same results and should not be used interchangeably. Total chlorophyll was determined for samples of canola seed and commercial canola oil by recognized spectrophoto- metric methods and by high-performance liquid chroma- tography (HPLC). The HPLC method, which summed all chlorophyll-related pigments detected, found approxi- mately 1.4 times more total chlorophyll per sample than did the spectrophotometric methods. The spectrophoto- metric methods are calibrated with only chlorophyll a and underestimate other chlorophyll pigments, which have lower extinction coefficients and different absorption max- ima. The HPLC method detects each pigment at its ab- sorption maxima and applies the appropriate absorptivity factor. Care must be taken when comparing results ob- tained by different methods. There appears to be a need for a standardized method of chlorophyll pigment measure- ment by HPLC.

KEY WORDS: Brassica, canola oil, canola seed, chlorophyll, chloro- phyll analysis, HPLC, pigments, spectrophotometer.

Cl~dorophyll measuxement in canola seed and oil is impor- tant to mouitor crop quality, particularly in years with early frost or for crops with late or uneven maturity. Several methods exist to measure total cb_lorophyll or individual chlorophyll components in canola seed and oil These methods include the spectrophotometric AOCS Official Method AK 2-92 (1), which is applicable to the measure- ment of total seed chlorophyll in rapeseed, measured as chlorophyll a (chl a); the spectrophotometric AOCS Official Method Cc 13d-55 (2), which is applicable to the detern-ina- tion of total chlorophyll, measured as chl a, in refined vegetable oils (although this has generally been applied to canola oil at all stages of processing); and recently published {3-5) high-performance liquid chromatography (HPLC) methods for chlorophyll pigment measurement in canola oil. These methods rely on the separation of chlorophyll pig- ments on a reversed-phase HPLC column and detection with either an ultra,Aolet (uv)/visible or a fluorescence detector. These methods are calibrated with samples of each chloro- phyll pigment detected. Different methods often will not yield the same results, so care must be taken when these results are compared. We have compared the results of chlorophyll analysis by two commonly used spectrophoto- metric methods for canola seed and oil samples, with results from HPLC by an unofficial but established method (4).

MATERIALS AND METHODS

Chlorophyll measurement in commercially extracted canola oil. Freshly extracted canola oil samples, including

*To whom correspondence should be addressed.

pressed, solvent-extracted and degummed oils, were ob- tained from a western Canadian canola crushing plant. Crude oil could not be obtained directly from the pro- cessor, and was prepared in the lab as a 50:50 mix of pressed and solvent-extracted otis. The fresh oil samples were placed in plastic bottles in a cooler and taken to the laboratory for immediate analysis. Samples were then stored for up to one month at room temperature in both light and dark, in a refrigerator and in a freezer.

HPLC technique. Oil was dissolved in acetone prior to analysis to give a solution of 25% oil. HPLC analysis was carried out according to the method of Endo et al. {4), ex- cept that the fluorescence detector was replaced with a photodiode array detector. The HPLC system consisted of two Waters {Waters Associates, Milford, MA) model 510 pumps, a Waters model 715 Ultra "Wisp sample processor and a "Waters model 994 programmable photodiode array detector. The column was stainless steel (220 mm X 4.6 mm) packed with O.D.S. 5 ~m {Pierce Chemical Co., Rockford, IL). Each set of oil samples {fresh and after storage} was run on the HPLC by using a 50-~L injection volume and a run time of 30 min, which was sufficient to allow all of the chlorophyll derivatives to elute. The mobile phase was water/methanol/acetone (4:36:60} at a flow rate of 1.0 mL/min. The photodiode array detector was used to scan peaks to identify the chlorophyll pig- ments by their characteristic absorption maxima. Wave- lengths used for quantitation were 642, 655, 662 and 667 nm. Chlorophyll pigments were identified from their ab- sorption spectra and retention times by comparison with standards of known composition (4).

Chl a was purchased from Fluka Chemical Co. (Ronkon- koma, NY), and chlorophyll b (Chl b) was purchased from Sigma Chemical Co. (St. Louis, MO). Pheophytins a and b (phy a and phy b) were prepared from chl a and b, respec- tively, by reaction with HC1 (6). Calibration curves were prepared for chl a, chl b, phy a and phy b. All curves were linear, with r 2 values ranging from 0.95-0.98. Endo et aL (4) had previously shown that these four compounds, plus pyropheophytin a, were the major pigments that occur in commercially-extracted canola oil.

Each set of samples included standards of chl b, chl a, b and phy a, and each day's run was quantitated by us- ing the standards from the same run. Standards were not

prepared for pheophorbide a, methylpheophorbide a or pyropheophytin a, so these were quantitated based on the phy a standard after multiplying by the ratio of the ex- tinction coefficients (i.e., 1.24 for pheophorbide a, 1.33 for methylpheophorbide a and 1.10 for pyropheophytin a). Total chlorophyll for each oil sample was calculated as the sum of all chlorophyll derivatives detected.

Spectrophotometric method. The spectrophotometric method used was the AOCS Official Method Cc 13d-55 after recalibration with isooctane/ethanol instead of methylene chloride (2). Oil samples were dissolved in a 3:1 mixture of isooctane/ethanol to give a solution of 10% oil.

Copyright �9 1994 by AOCS Press JAOCS, Vol. 71, no. 9 (September 1994)

932

K. WARD ET AL.

Samples were filtered, and the absorbance was measured at 625, 665 and 705 nm in a 1-cm cuvette with a Varian (Palo Alto, CA) DMS 200 UV-Visible spectrophotometer. Two measurements were made on each sample, and the results were averaged.

Chlorophyll measurement in canola seeds. Canola seed was harvested at various stages of matur i ty throughout the ripening period to span a wide range of chlorophyll levels. These seed samples were analyzed by extract ing the chlorophyll and measuring the absorbance with a Varian DMS 200 UV-Visible spectrophotometer, according to the AOCS Official Method AK 2-92 (1). One-gram sam- ples of freeze-dried seed were weighed out and placed in stainless steel extract ion tubes with three ball bearings and 30 mL of 3:1 isooctane/ethanol. Samples were shaken for one hour, filtered, dilutions were prepared if required, and the absorbance readings were measured. Three wave- lengths were used (625.5, 665.5 and 705.5 nm) to measure the absorption peak for chlorophyll with corrections on either side. Two extractions and measurements were made on each sample, and the results were averaged. The absorb- ance readings were converted to chlorophyll levels by means of the following formula, which was developed from standards of known chlorophyll concentration:

C.O.D. = Abs. (665.5 rim) -- labs. (625.5 nm)+ Abs. (705.5 rim)i/2 Chl (mg kg -1) = 390 * C.O.D. [1]

where Abs. = absorbance; C.O.D. -- corrected optical density.

To extract and prepare the samples for HPLC analysis, 2-g samples of freeze-dried seed were weighed out and placed in stainless-steel extract ion tubes with three ball bearings and 30 mL of 3:1 isooctane/ethanol. Samples were shaken for i h and filtered, and the oil extract containing the chlorophyll was collected. High-chlorophyll samples were diluted if necessary. Samples were evaporated to dryness under nitrogen and resuspended in acetone. HPLC analysis was carried out as described for the oil samples, with an injection volume of between 10-50 t~L, depending on the concentrat ion of chlorophyll pigments in the sample.

RESULTS AND DISCUSSION

Chlorophyll measurement in commercially extracted canola oil. When total chlorophyll levels were measured by the spectrophotometric method for each oil sample, a small apparent decrease in total chlorophyll was observed in the pressed, solvent-extracted and crude oils during storage for one month (Figs. 1-3). The apparent decrease was greatest for the oils stored at room temperature in the light, followed by storage at room temperature in the dark, in the refrigerator and in the freezer, respectively. On the other hand, there was no apparent decrease in total chlorophyll during oil storage under any of the conditions t e s t ed in t h e degummed oils (Fig. 4).

Previous studies have never indicated a decrease in total chlorophyll during oil storage. In another paper (7), we showed tha t during oil storage chl b was converted to phy b, and chl a was converted to phy a and pyropheophyt in a. This conversion occurred most quickly in oils stored at room temperature in the light, followed by storage at room temperature in the dark and in refrigerated storage, respec-

40

50

20

~A

Storage Condit ions O - - O Freezer �9 - - �9 Ref r igera tor A - - A Dark Bench A - - A Light Bench

I I - - 1 7 1 4 21 28

Days of Storage

FIG. 1. Total chlorophyll in pressed oil during storage as measured by spectrophotometry.

o o

8

so /

40

50

Storage Condit ions O O Freezer �9 - - �9 Ref r igerator ~ - - A Dark Bench A - - A Light Bench

I I - - I I

7 1 4 21 28

Days of Storage

FIG. 2. Total chlorophyll in solvent-extracted oil during storage as measured by spectrophotometry.

(3_ o o .c c3

6O

4 0 -

- A

Storage Condit ions O - - O Freezer �9 �9 Ref r igerator z ~ - - - z& Dark Bench A - - A Light Bench

30 - - - - I ........ l ........ I . . . . . 1 - - 0 7 t 4 21 28

Days of Storage

FIG. 3. Total chlorophyll in crude oil during storage as measured by spectrophotometry.

tively. Only minor changes occurred in oil samples tha t were frozen.

Therefore, the apparent decrease in total chlorophyll during storage, as measured by the spectrophotometric method, was not a real decrease. During storage, chloro- phylls were converted to pheophytins and some pyropheo- phytins. The spectrophotometr ic method of chlorophyll

JAOCS, Vol. 71, no. 9 (September 1994)

A COMPARISON OF HPLC AND SPECTROPHOTOMETRY

933

6O

o o 5 4 0 Storage Conditions

O - - 0 Freezer Q - - �9 Refrigerator A - - A D a r k Bench A - - A Ligh~ Bench

50 i t ~- . . . . . t 0 7 t4 21 28

Days of Storage

FIG. 4. Total chlorophyll in degummed oil during storage as measured by spectrophotometry.

measurement is calibrated with only chl a, which has a milch higher ext inct ion coefficient (at 665 nm) than phy a or any of the other chlorophyll derivatives (Table 1). Thus, pheophytins and pyropheophyt ins in the stored oil samples, al though present at the same concentrat ion as the original chlorophylls, will produce a lower absorbance reading on the spectrophotometer, leading to an under- est imate of total chlorophyll. In the pressed, solvent- extracted and crude oils, the conversion of chl a to phy a and pyropheophytin a was responsible for the apparent decrease in total chlorophyll observed. The degummed oil contained mainly pheophytins and pyropheophytins to begin with, so little conversion was possibl~ and there was no decrease in absorbance.

Suzuki and Nishioka (5) also discussed this apparent discrepancy between pigment concentrations when mea- sured by the AOCS spectrophotometric method Cc 13d-55 and by an HPLC method. They detected concentrations of pheophytins and pyropheophytins tha t were 1.4 times higher by HPLC than by the spectrophotometric method. They also s tated tha t the calibration of the spectrophoto- metric method with chl a accounted for this observation.

In our study, total chlorophyll, as determined by HPLC and calculated as the sum of all chlorophyll derivatives detected, did not agree with total chlorophyll as measured by the spectrophtornetric method. There was a s trong positive correlation between total chlorophyll levels deter- mined by HPLC and spect rophotometry (r = 0.73, P > [r t = 0.0001). The mean ratio of HPLC chlorophyll/spectro-

photometer chlorophyll was 1.40 with a s tandard devia- tion of 0.27. Therefore, the HPLC method detected, on average, 1.4 t imes more total chlorophyll per sample than did the spectrophotometr ic method. However, this value varied a great deal between samples.

Chlorophyll measurement in canola seeds. The relation- ship between total chlorophyll as measured by HPLC and calculated as the sum of all chlorophyll derivatives de- tected in a seed sample, and total chlorophyll, measured by the AOCS spectrophotometric method AK 2-92 (1), was compared for seed samples ranging from green to fully ripe, to cover a wide range of chlorophyll levels.

The HPLC method detected, on average, 1.4 times more total chlorophyll than the spectrophotometr ic method. This value of 1.4 was identical to tha t found for the com- mercially-extracted oil samples in the previous section. I t also agrees with the results of Suzuki and Nishioka (5). There was a high positive correlation (+0.93) between total chlorophyll as measured by HPLC and spectrophoto- metry, significant at the P = .0001 level.

The ratio between the HPLC measurement of total chlorophyll p igments and the spec t ropho tomet r i c readings varied between 0.3-3.0 for individual samples. Over the entire study, the ratio of HPLC/spectrophoto- meter chlorophyll was 1.37 with a s tandard deviation of 0.43. Any samples tha t contained less than 3.0 mg - kg -1 total chlorophyll, as measured by the spectrophotometric method, were not included in the calculations. The spec- t rophotometr ic method only detects chlorophyll to an ac- curacy of within 3 mg �9 kg-t; therefore, samples contain- ing lower levels of chlorophyll are below the limits of ac- curate detection, and the ratio for these low pigment levels could be abnormally high or low by chance alone. Once again, the discrepancy in total chlorophyll, as measured by HPLC and spectrophotometry, is explained by the calibration of the spectrophotometr ic method in which only chl a is used, leading to an underest imation of any other pigments present.

These results i l lustrate a potential problem in chloro- phyll analysis studies. For any given study, the san-re method of chlorophyll measurement mus t be used throughout the entire study, as results from different methods cannot be directly compared. Also, the results of separate studies, which have been carried out with dif- ferent methods of chlorophyll measurement, should not be directly compared. The spectrophotometric and HPLC methods of chlorophyll measurement were not designed for the same use. The spectrophotometr ic method gives a relatively rapid measurement of total chlorophyll and was designed to compare samples of the same type. The

TABLE 1

Absorption Characteristics of Chlorophylls and Related Pigments

Pigment Maximum ~ E a Maximum ~. E a Reference

Pheophorbide a 409 119200 667 55200 8 Methylpheophorbide a 408.5 122500 667 59200 9 Chlorophyll b 455 131000 645 47100 10 Chlorophyll a 430 94700 663 75000 10 Pheophytin b 434.5 145000 654 27800 8 Pheophytin a 409 101800 666 44500 11 Pyropheophytin a 409 102400 667 49000 9 aE, Absorbtivity (molar extinction coefficient) in acetone (except methylpheophorbide a in ether).

JAOCS, Vol. 71, no. 9 (September 1994)

934

K. WARD E T AL.

more d iverse t he s a m p l e s t h a t a re b e i n g compared , t h e g r e a t e r t h e degree of e r ro r t h a t m a y be in t roduced . Fo r example , a h i g h - q u a l i t y seed s a m p l e t h a t c o n t a i n e d 15 m g �9 k g -1 ch lo rophy l l a would p r o d u c e a h ighe r absorb- ance r e a d i n g t h a n a d a m a g e d seed s a m p l e t h a t con ta ined 15 m g �9 k g -1 p h y a.

The H P L C me thod , on t h e o t h e r hand , can accu ra t e ly d e t e r m i n e the levels of each ch lorophyl l p i g m e n t in a sam- ple. Therefore, i t can be u s e d to m a k e c o m p a r i s o n s be- tween d i f fe ren t t y p e s of samples , I t is, however, m u c h more t i m e - c o n s u m i n g a n d expens ive t h a n the spec t ro- p h o t o m e t r i c m e t h o d if one is on ly i n t e r e s t ed in t o t a l ch lorophyl l . A s more l a b o r a t o r i e s p u r c h a s e H P L C equip- ment , a s t anda rd i zed H P L C m e t h o d for chlorophyl l deter- m i n a t i o n in cano la seed and oil wou ld be useful .

ACNOWLEDGMENTS Financial support was provided by the Canadian Wheat Board and the University of Manitoba. Staff at the Grain Research Lab of the Carmdiarl Grain Commission provided materials, equipment and assistance. Oil samples were generously provided by CanAmera Foods. Contribution numbers for this paper are 963 from the Univer-

sity of Manitoba, Department of Plant Science, and 715 from the Canadian Grain Commission.

REFERENCES 1. Official Methods and Recommended Practices of the American

Oil Chemists ' Society, 3rd edn., edited by V.C. Mehlenbacher, T.H. Hopper, E.M. Smallee, W.E. Link, R.O. Walker and R.C. Walker, American Oil Chemists' Society, Champaign, Method Ak 2-92, 1992.

2. Ibid., AOCS Method Cc 13d-55, 1989. 3. Daun, J.K., and C.T. Thoresteinson, J. Am. Oil Chem. Soa 66.'1124

(1989). 4. Endo, Y., C.T. Thorsteinson and J.K. Daun, Ibid. 69:564 (1992). 5. Suzuki, K., and A. Nishioka, Ibid. 70:837 (1993). 6. Hyanninen, P., and N. Ellfolk, Acta Chem. ScandL 27.'1463 (1973). 7. Ward, K., R. Scarth, J.K. Daun and C.T. Thoresteinson, J. Am.

Oil Chem. Soc. 71:811 (1994). 8. Wasielewski, M.R., and W.A. Svec, J. Org. Chem. 45".1969 (1980}. 9. Pennington, F.C., H.H. Strain, W.A. Svec and J.J. Katz, J. Am.

Chem. Soc. 86:1418 (1964). 10. MacKinney, G., J. Biol. Chem. 132:91 {1940). 11. Wilson, J.R., M.-D. Nutting and G.F. Bailey, Anal. Chem. 34:1331

(1962).

[Received January 26, 1994; accepted April 22, 1994]

JAOCS, VoL 71, no. 9 (September 1994)